Big Data analytics for improved prediction of ligand binding and conformational selection.

Area of Science:

• Computational biology
• Biophysics
• Machine learning

Background:

• Understanding protein:ligand interactions is crucial for drug discovery.
• The protein conformational selection mechanism governs how proteins bind to ligands.
• Predicting binding events requires analyzing numerous protein conformations.

Purpose of the Study:

• To introduce novel Big Data analytics techniques for studying the protein conformational selection mechanism.
• To enhance the prediction accuracy of protein:ligand binding.
• To identify key protein properties influencing binding probability.

Main Methods:

• Application of machine learning and deep learning approaches (Big Data analytics).
• Efficient processing of large datasets of protein:ligand complexes.
• Analysis of specific G protein-coupled receptor (GPCR) proteins: ADORA2A, ADRB2, OPRD1, and OPRK1.

Main Results:

• Big Data analytics significantly enhanced the prediction of binding protein conformations (10-38 times better than random selection).
• The techniques enable efficient processing of vast ensembles of protein conformations.
• Identification of physico-chemical features predictive of binding conformations, with some shared and some protein-specific properties.

Conclusions:

• Big Data analytics offers a powerful approach to understanding the protein conformational selection mechanism.
• This method facilitates the systematic identification of protein "binding conformations".
• Biophysical properties driving conformation selection show both shared and protein-specific characteristics.